Varos Glossary

Bias Mitigation

Even the most sophisticated artificial intelligences may be subject to the biases of their creators. At best, these algorithmic biases can lead to inaccurate or nonsensical outputs. At worst, they can cause real and lasting harm.. 

If artificial intelligence and machine learning are ever to be used responsibly and ethically, we must first find a way to mitigate bias.

What Causes Bias in AI? 

There are many different types of bias — but within the context of machine learning, these biases are all caused by one or more of the following issues: 

  • The AI is designed to support or replace an inherently biased process or system.
  • The AI is trained based on data that is itself biased, either by its nature or because it inadequately represents certain subgroups. 
  • The AI's creators have, knowingly or unknowingly, made biased assumptions about the system. 
  • The AI uses variables which, by proxy, create bias — for instance, a zip code may be a proxy for a subject's race.

In most cases, these biases are unintentional. Unfortunately, while biases in human decision-making may have small impacts, even a small bias in an algorithm may lead to exponentially worse decision-making. This is because an AI doesn't simply learn the biases it's taught — it tends to amplify them. 

What is Bias Mitigation? 

Bias mitigation is the process of identifying and eliminating the biases that may be present in an artificial intelligence. This may be accomplished manually or through bias mitigation algorithms. Mitigation techniques can be divided into several broad categories, and may be either manual or algorithm-based. 

Primary Bias Mitigation Strategies

Manual Bias Mitigation

Manual bias mitigation techniques include: 

  • Creating a more balanced dataset by obtaining data from multiple sources or simply spending more time collecting training data
  • Adjusting the parameters that determine how the machine learning model parses its training data. 
  • Removing or reweighing certain features or variables.

Unfortunately, none of these techniques are necessarily effective at addressing unconscious biases. For this reason, it's generally advisable to also incorporate algorithmic bias mitigation strategies. 

Algorithmic Bias Mitigation

Bias mitigation for machine learning can be divided into three broad categories based on where in the machine learning model's life cycle it is applied. 

Pre-Processing Bias Mitigation

Techniques such as data preprocessing assess training data for potential biases before it's fed into a machine learning model, with the goal of ensuring said data is as fair and impartial as possible. Unfortunately, this does little to address the biases that may be present in the model itself. 

In-Processing Bias Mitigation

In-processing bias mitigation occurs during training, and generally comes in one of two forms. Fairness-aware algorithms are explicitly designed to be impartial, with checks and balances to ensure that they don't favor or discriminate against any group. Adversarial debiasing, meanwhile, trains the machine learning model alongside an adversary designed to identify and mitigate biased decision-making. 

Post-Processing Bias Mitigation

After a model has been trained, post-processing bias mitigation may be applied to ensure it remains impartial. This form of bias mitigation works in one of two ways. Either it leverages probabilities to change outputs, or it identifies groups experiencing bias and projects favorable outcomes onto them.

While this technically accomplishes the objective of mitigating bias, it also impacts the accuracy of the system. As such, it may not necessarily represent an ideal solution to the issue of bias.